flowgentic.langGraph.execution_wrappers¶

execution_wrappers ¶

LangGraph/AsyncFlow Integration: bridge AsyncFlow tasks and LangChain tools with built-in retries, backoff, and timeouts.

Key features: - Define AsyncFlow tasks with @flow.function_task - Expose as LangChain tools via @integration.asyncflow_tool(...) - Sensible defaults for fault tolerance (no config required)

TBD

- block for executing the agents in parallel!¶

Classes:

Name	Description
`AsyncFlowType`	Enum defining the flow_type of AsyncFlow decoration
`ExecutionWrappersLangraph`	Enhanced integration between AsyncFlow WorkflowEngine and LangChain tools.

AsyncFlowType ¶

Bases: Enum

Enum defining the flow_type of AsyncFlow decoration

ExecutionWrappersLangraph ¶

ExecutionWrappersLangraph(flow: WorkflowEngine, instrospector: GraphIntrospector)

Enhanced integration between AsyncFlow WorkflowEngine and LangChain tools.

Supports both traditional workflow patterns and React agent orchestration with supervisor patterns and parallel execution.

Methods:

Name	Description
`asyncflow`	Unified decorator to register async functions as AsyncFlow tasks with different behaviors.

asyncflow ¶

asyncflow(func: Optional[Callable] = None, *, flow_type: AsyncFlowType = None, retry: Optional[RetryConfig] = None, **kwargs) -> Callable

Unified decorator to register async functions as AsyncFlow tasks with different behaviors.

Parameters:

Name	Type	Description	Default
`flow_type`	`AsyncFlowType`	AsyncFlowType enum specifying the decoration behavior: - AGENT_TOOL_AS_FUNCTION: Creates a LangChain tool (with @tool wrapper) - AGENT_TOOL_AS_SERVICE: Creates a LangChain tool with persistent service instance - FUNCTION_TASK: Creates a simple asyncflow task (with args, *kwargs) - SERVICE_TASK: Creates a persistent service task - EXECUTION_BLOCK: Creates a block node	`None`
`retry`	`Optional[RetryConfig]`	Optional retry configuration	`None`